The proposed research has two components. The first is the construction of a unique, very large, representative data set matching workers to employers, using information on workers from the 2000 Decennial Long-Form of the Census. The matching technology we will use is one we are developing and refining to construct a similar data set using currently available 1990 data, where the employer characteristics in these data sets are obtained in part from information on multiple workers working for each employer. For each individual worker, the data sets include detailed information on both residential location and geographic location of the place of work. The second component of the research program is to use the 1990 and 2000 data to study the interdependence of residential and labor market segregation and the effects of segregation on economic outcomes. [unreadable] [unreadable]